Inhibition of DNA primase and polymerase. alpha. by

requires a 3'-hydroxyl on the incoming NTP in order to polymerize it, while the ... analyzed. Removing the 3'-hydroxyl of the incoming triphosphate de...
1 downloads 0 Views 2MB Size
Download PDF
4720

Biochemistry 1992, 31, 4720-4728

Inhibition of DNA Primase and Polymerase a by Arabinofuranosylnucleoside Triphosphates and Related Compoundst Robert D. Kuchta,* Diane Ilsley, Kasey D. Kravig, Sybilla Schubert, and Bristol Harris Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-021 5 Received January 10, 1992; Revised Manuscript Received February 25, 1992

ABSTRACT: Inhibition of DNA primase and polymerase a from calf thymus was examined. DNA primase requires a 3’-hydroxyl on the incoming NTP in order to polymerize it, while the 2’-hydroxyl is advantageous, but not essential. Amazingly, primase prefers to polymerize araATP rather than ATP by 4-fold (kat/KM). However, after incorporation of an araNMP into the growing primer, further synthesis is abolished. The 2’- and 3’-hydroxyls of the incoming nucleotide appear relatively unimportant for nucleotide binding to primase. Polymerization of nucleoside triphosphates by DNA polymerase CY onto a DNA primer was similarly analyzed. Removing the 3’-hydroxyl of the incoming triphosphate decreases the polymerization rate >1000-fold (kcat/KM),while a 2’-hydroxyl in the rib0 configuration abolishes polymerization. If the 2’hydroxyl is in the ara configuration, there is almost no effect on polymerization. An araCMP or ddCMP at the 3’-terminus of a DNA primer slightly decreased DNA binding as well as binding of the next correct 2’-dNTP. Changing the primer from DNA to RNA dramatically and unpredictably altered the interactions of pol a with araNTPs and ddNTPs. Compared to the identical DNA primer, pol a discriminated 4-fold better against araCTP polymerization when the primer was RNA, but 85-fold wome against ddCTP polymerization. Additionally, pol a elongated RNA primers containing 3’-terminal araNMPs more efficiently than the identical DNA substrate.

A

variety of nucleotide analogues inhibit DNA synthesis and are clinically useful. These include the arabinofuranosyl nucleosides arabinofuranosylcytosine(araC),’ which is used to treat certain leukemias, and araA, which exhibits antiviral properties (Weil et al., 1980; Keeney & Buchanan, 1975). In each case, the active form of the compound is thought to be the triphosphate. Both compounds inhibit DNA replication in vivo, and this is a likely source of their cytotoxicity. AraA and araC are readily incorporated into DNA during replication, primarily at internucleotide linkages, but also at the 3’-terminus (Major et al., 1981, 1982). As the triphosphate, they potently inhibit several enzymes involved in DNA replication, including the DNA polymerases and DNA primase (Dicioccio & Srivastava, 1977; Lee et al., 1980; Kuchta & Willhelm, 1991). Thus, the araNTPs could potentially disrupt DNA replication by inhibiting initiation of new replicons (primase inhibition), by simple competitive inhibition with respect to a 2’-dNTP (polymerase inhibition), or via incorporation into the DNA and a large reduction in the rate of polymerization of the next 2’-dNTP (polymerase inhibition). DNA primase and pol a are essential for DNA replication. The two activities copurify as a four-peptide complex (Kaguni et al., 1983; Brooks & Dumas, 1989). On single-stranded DNA, primase synthesizes short RNA oligomers that pol a then elongates via 2’-dNTP polymerization. While primase synthesizes products 2-10 nucleotides long, pol a only utilizes those primers at least 7 nucleotides long, both in vivo and in vitro (Kuchta et al., 1990; Kitani et al., 1984; Hay et al., 1984). Previously, we found that araATP is a more potent inhibitor of primase than of pol a. AraATP is readily polymerized into ‘This work was supported by grants from the National Science Foundation (DMB9105808), the Council for Tobacco Research (2612), and the UROP program at the University of Colorado. R.D.K. was partially supported by a Junior Faculty Research Award from the American Cancer Society (337). * Address correspondence to this author.

primers and results in increased synthesis of short primers that pol a cannot elongate. Pol CY polymerizes dNTPs onto DNA primers containing an araNMP at the 3’-terminw much slower than if the primer contains a 2’-dNMP at the 3’4erminus (Mikita & Beardsley, 1988). Curiously, however, we observed that pol a polymerizes multiple araAMPs onto primase-synthesized RNA primers. These data raise the possibility that inhibitors may interact differently with pol a when the primer is RNA. In addition to the araNTPs, several clinically useful nucleotide analogues contain modifications in the 2’- and 3’hydroxyls. Included in this group are 3’-azido-3’-deoxythymidine and the dideoxy nucleotides, which exhibit potent anti-HIV activity (Larder et al., 1989). In these experiments, we further examine primase inhibition by araATP and related nucleotides, and also examine pol a inhibition. They provide insights into the role of the 2’- and 3’-hydroxyls during nucleotide binding and polymerization by each enzyme and show that RNA primers greatly modulate pol a inhibition by nucleotide analogues relative to DNA primers. EXPERIMENTAL PROCEDURES Materials Pol a was purified from calf thymus using immunoaffinity chromatography as described previously (Kuchta & Willhelm, 1991). One unit of pol a polymerizes 1 nmol of 2’dNTP h-’ on activated calf thymus DNA at 37 OC, and 1 unit of primase polymerizes 1 nmol of ATP h-’ on a poly(dT) template. [3H]AraAMPand [3H]dATPwere from Moravek Biochemicals, while all other radiolabeled compounds were from ICN. 3’-AraAMP was generously provided by Wayne Miller at Burroughs-Wellcome Corp. All chemicals were reagent grade I Abbreviations: araA, 9-~-~arabinofuranosyladenosine; araC, arabinofuranosylcytosine; ddNTP, 2’,3’-dideoxynucleoside triphosphate; EDTA, ethylenediaminetetraaceticacid (Na salt); pol a, DNA polymerase a:primase; Tris, tris(hydroxylmethy1)aminomethane (HCl salt).

0006-2960/92/043 1-4720%03.00/0 0 1992 American Chemical Society

Inhibition of DNA Primase and Pol a by Arabinofuranosylnucleoside Chart I: Synthetic Primer-TemplatesUsed DNAG

TCC ATA TCA CAT (3’) AGG TAT AGT GTA GAT C l l ATC ATC T

DNA42

GCG CCG AAA (3’) ll-i CGC GGC T l l GTAAAA GTCCCC GTAAAA GTCCCC GTAAAA

RNA421

GCG CCG AAA (3’)

l l l CGC GGC Tll GTAAAA GTCCCC GTAAAA GTCCCC GTAAAA

”The primer strand is all RNA, while the DNA strand is all DNA.

or better. DNA oligonucleotides of defined sequence were obtained from Oligos, Etc., while the RNA primer was synthesized on an Applied Biosystems DNA synthesizer. Duplexes were formed as described previously (Kuchta et al., duplex 1987), and are listed in Chart I. (dT)25-30*(rA)12-18 was annealed in a 2:l base ratio. DN& containing a ddCMP polymerized onto the primer 3’-terminus was synthesized using Klenow fragment [(exo-), kindly provided by Dr. C. Joyce] and a 20-fold excess of ddCTP. DNA, containing a 3’4erminal araCMP was synthesized similarly, except ddCTP was replaced with araCTP. DNAG containing a 3’-terminal 2’-dCMP was synthesized using pol a and a 2-fold excess of 2’-dCTP. DNA was purified by EtOH precipitation (Maniatis et al., 1982). Each DNA was 5’-32P-labeledand analyzed by gel electrophoresis followed by phosphorimagery to ascertain that all of the primer strand was elongated. Additionally, we added Klenow fragment (exo’) and four dNTPs to elongate each of the DNAs and show that all of the primer strand could be elongated (the exonuclease removes the 3’4erminal araCMP or ddCMP, and then the polymerase activity elongates it). Methods Unless noted, all assays contained 5 mM MgC12 and 50 mM Tris, pH 7.5, and were performed at 37 “C. Rates were measured under initial velocity conditions. Radioactivity incorporated into products was measured either by using a DE81 filter binding assay (Sheaff et al., 1991) or by polyacrylamide gel electrophoresis [ 18% acrylamide with the Tris/borate/ EDTA buffer increased 1.5-fold in the gel, but standard concentration in the running buffer (Sheaff et al., 1991)] followed by phosphorimagery on a Molecular Dynamics PhosphorImager. Pol a Activity. Pol a activity was typically measured in assays containing 50 pM poly(dT)sligo(rA) (20: 1 base ratio) and [ c ~ - ~ ~ P ] ~ and A T Pthe , amount of 32Pincorporated into DNA was measured. To obtain Kis, the data were analyzed using Dixon plots. To determine the relative k,,/K, for araCTP versus 2’-dCTP, assays contained 0.4 pM DN& (5’-32P-labeledin the primer strand), 10 pM each of 2’-dNTP, and various amounts of araCTP. Assays were allowed to proceed until all starting material was elongated and then quenched by the addition of 3 parts gel loading buffer. Products were separated by gel electrophoresis, and the amount of primer elongated by 1 nucleotide and by >1 nucleotide was quantified by phosphorimagery. Incorporation of araCTP into the DNA resulted in a product that is not readily elongated by pol a. Thus, the primer elongated by one nucleotide represents polymerization of araCTP, and longer products reflect polymerization of 2’-dCTP opposite the template G. The relative k,,/KM can then be obtained as the x intercept from a plot of (fraction of primers containing 2’-dCMP)-l versus [araCTP]/[dCTP] (Fersht, 1985). Background levels for quantitation of the primer elongated by one nucleotide were obtained from reactions containing all four dNTPs and no araCTP, and the background for longer polymerization

Biochemistry, Vol. 31, No. 19, 1992 4721

products was obtained from reactions where no dNTPs were present. Analogous procedures were used for determining the relative k a t / & values for ddCTP versus 2’-dCTP, and araATP or ddATP versus 2’-dATP. In measurements of araNTP polymerization, control experiments showed that 95% of the [3H] comigrated with authentic analogues as primase inhibitors in order to understand the role araATP. of the 2’- and 3’-hydroxyls. For rate measurements, we only Ratio of AraAMP to AMP in Primers. Primers were synthesized in assays (20 pL) containing 100 pM [CY-~~PIATP, considered those primers at least seven nucleotides long to be products since pol a can only elongate these primers (Kuchta 1.3 or 2.6 pM [3H]araATP, 50 pM poly(dT), 10 mM diet al., 1990), and the function of primase is to synthesize thiothreitol, 50 pM aphidicolin, and 4.5 pM DN&. Reactions primers that pol a can use. For each compound, competitive were quenched by the addition of 10 pL of 50 mM EDTA and inhibition with ATP was observed (Table I). Inhibition was 20 pL of H20. Primers were separated from remaining ATP due both to a decreased total number of primers and to a and araATP by chromatography on a 1-mL Sephadex G-25 decreased fraction of the primers reaching seven nucleotides column with 30% EtOH in H 2 0 as solvent. Primer-containing in length (not shown). Interestingly, even though primase fractions were pooled and lyophilized to dryness. The residue activity was always measured on a poly(dT) template, rewas resuspended in 10 pL of H20 and 30 p L of gel loading placing the adenosine compound with the corresponding cybuffer. Primers were separated by gel electrophoresis (22% tosine analogue had relatively little effect on the potency of acrylamide) and located by autoradiography. Bands (0.5-1 .O inhibition, particularly if the analogue was not polymerized cm2) containing each primer were excised from the gel, and (3’-dATP versus 3’-dCTP, and ddATP versus ddCTP) , Re1 mL of H 2 0 added. To speed elution of the primers from moval of either the 2’- or the 3’-hydroxyl decreased inhibition; the gel, samples were subjected to two freeze-thaw cycles. hence, both the 2’- and 3’-hydroxyls are important for interScintillation fluid (1 0 mL, National Diagnostics Econoscint) acting with primase. was added, and the amount of 3H and 32Pin each sample was For those nucleotides that pol a also utilized,2 we needed determined by scintillation counting (Packard 1600TR scinto rule out the possibility that observed inhibition of primase tillation counter). was due to interactions at the polymerase active site. We Location of AraAMP and 2‘-dAMP in Primers. Primers previously showed that addition of DNA, and aphidicolin to were synthesized and purified from reactions identical to those pol a inhibits pol a activity via formation of a tightly bound described above for determining the ratio of araAMP/AMP pol a*DN&aphidicolin ternary complex (Sheaff et al., 1991). in primers, except that the 32Pwas 80 cpm pmol-’. Primers Primase inhibition by araATP and ddATP, which are polymwere hydrolyzed to 3’-NMPs and nucleosides by treatment erized onto primers by pol a,was not altered when pol a was with 0.5 unit of spleen phosphodiesterase and 4 units of RNase inhibited with aphidicolin and DNAG. Since pol a rapidly T2 (Sigma) for 30 or 60 min at 37 OC. Products were sepaelongated primers with 2’-dATP, the Kifor 2’-dATP in Table rated by TLC on silica gel with a solvent of iPrOH/ I was only measured with pol CY inhibited. NH40H/H20(63:30:7), and the amount of radioactivity was Polymerization of Nucleotides by Primase. In addition to determined by cutting the TLC into small pieces and adding ATP, both araATP and 2‘-dATP were incorporated into 0.5 mL, of H 2 0 followed by 5 mL of scintillation fluid. Similar primers (Table I and Figure 1). Incorporation of each anaprocedures were used to examine 2I-dAMP incorporation into logue resulted in products of altered electrophoretic mobility. primers, except reactions contained [3H]2’-dATP instead of Inhibiting pol a with DN& and aphidicolin did not prevent [3H] araATP. incorporation of the analogues into primers, indicating that Binding of DNA and Nucleotides to Pol a. The KD value primase was in fact responsible for their polymerization. of DNA was measured as a Kias described previously (Sheaff 3’-dATP and ddATP were not polymerized into primers et al., 1991). The Kd values for nucleotide binding to pol asDNA complexes were measured using the equilibrium inPol a activity specifically refers to the DNA polymerase activity. hibition assay and associated kinetics (Sheaff et al., 1991). ~

Inhibition of DNA Primase and Pol a! by Arabinofuranosylnucleoside

Biochemistry, Vol. 31, No. 19, 1992 4723

Table 11: Relative kat/& for Polymerization of AraATP versus ATP bv Primase

9 ‘Primer length refers to the length of the primer to which either araATP or ATP is beina Dolvmerized.

abcdefghi jklm - a 2’ a d 3 ’ r a A T P

d A T P

r a C T P

d A T P

d A T P

FIGURE1 : Polymerizationof NTPs by primase. The length of primers is noted next to lane m. Assays contained pol aeprimase, 50 p M poly(dT), 100 p M [CY-~*P]ATP, and nothing (lanes b and c), 5 p M araATP (lanes d and e), 20 p M 2’-dATP (lanes f and g), 50 p M araCTP (lanes h and i), 100 pM ddATP (lanes j and k), or 50 p M 3’-dATP (lanes 1 and m). The assay in lane a lacked primase. Reactions for lanes c, e, g, i, k, and m also contained 100 pM aphidicolin and 7 pM DN& to inhibit pol a . In lane f, the absence of primers 7-10 nucleotides long is due to their elongation by pol a .

(Figure 1). When 3’-dATP was added to primase assays, new products of slightly lower mobility and corresponding to each of the 7-1 0-nucleotide-long primers were observed, indicating that 3’-dAMP was incorporated into products (lane 1). However, there were no corresponding products for the 2-6nucleotidelong primers, suggesting that pol a! polymerized the 3’-dAMP onto the primers. This was confirmed by including DNA, and aphidicolin to inhibit pol a! (lane m), which eliminated these new products. Similarly, pol a! polymerized ddATP onto primers; however, there was no detectable polymerization of ddATP by primase (lanes j and k). With a poly(dT) template, there was no detectable polymerization of CTP, 2’-dCTP, 3’-dCTP, or ddCTP. This was shown by (i) no observed products of altered electrophoretic mobility when these analogues were included in assays, (ii) no detectable incorporation into primers of 32P from [a32P]CTPor [cx-~~P] 2’-dCTP and, (iii) in assays containing [cu-~~PIATP and either 3’-dCTP or ddCTP, elongation of all primers 7-10 nucleotides long by pol a! upon addition of 2‘dATP. Incorporation of 3’-dCTP or ddCTP would have re-

sulted in a species that could not be elongated. Surprisingly, however, primase incorporated araCTP into primers as demonstrated by the appearance of products of altered electrophoretic mobility (Figure 1, lanes h and i). In order to test for pol a! catalyzed elongation of araC-containing primers, primers were fmt synthesized in assays containing [cu-~?]ATP and araCTP, and then 2’-dATP was added. While primers of normal electrophoretic mobility were rapidly elongated, pol a! was unable to elongate the araC-containing primers (not shown). Bimase Prefers To Polymerize AraATP Rather than ATP. The selectivity for polymerization of araATP vs ATP by primase was measured. Incorporation of an araAMP into primers alters their electrophoretic mobility and, as will be shown below, results in chain termination. Thus, we can determine how frequently primase polymerizes araATP or ATP onto a given length primer as a function of [araATP] / [ATP], and hence selectivity. The relative ( k , , / K M ) , ~ ~ / ( k ~ ~ / K Mfor ) * polymerization ~ onto primers 2-8 nucleotides long is shown in Table 11. Pol a! can polymerize nucleotides onto primers 1 7 nucleotides long; thus, the significantly higher kat/KM ratios for the longer primers may reflect small amounts of araATP polymerization by pol a!, even though assays contained 200 pM aphidicolin and 5 pM D N b . From the shorter primers, however, primase clearly prefers to polymerize araATP rather than ATP. Location of AraATP and 2‘-dATP in Primers. Primase readily polymerized araATP; hence, we determined where in primers the araAMP was located. The number of araAMPs incorporated into each length primer was quantified by synthesizing primers from [CU-~~PIATP and [3H]araATP. Primers were separated by gel electrophoresis, and the amount of 3H and 32Pin each length product was determined. Importantly, 50 pM aphidicolin and 4.5 pM D N b were included to inhibit pol a!. Primers 6-9 nucleotides long were analyzed. With 2.6 p M araATP,those primers that migrated identically to primers formed in the absence of araATP contained 0.08 f 0.04 araAMP per primer, while primers of altered electrophoretic mobility contained 0.89 f 0.1 araAMP per primer. Thus, incorporation of araAMP into a primer alters the electrophoretic mobility, and primase incorporates only one araAMP per primer. Similar data were obtained with 1.3 p M araATP-primers of “normal” electrophoretic mobility contained 0.03 i 0.01 araAMP per primer, while those of altered electrophoretic mobility contained 0.7 f 0.1 araAMP per primer. The presence of one araA per primer suggested that it would be located at either the 5’- or the 3’-terminus. Only the 5‘terminal nucleotide of the primers retains the y-phosphate of the NTP; hence, we used [y-32P]araATPto demonstrate that araATP was not incorporated at this position. In assays containing 100 pM ATP and 6.2 pM [ ~ - ~ ~ p ] a r a A 20000 2’-dCTP vs araCTP 1.o 2.0 9.2 2’dCTP vs ddCTP 5000 1300 15 2‘-dCTP vs C T P >20000 3.1 3.1 2’-dCTP vs araCTPb 4900 >3000 2‘-dCTP vs ddCTPb “This measurement is for the first requirement for 2‘-dATP or 2’dCTP polymerization (noted with an asterisk in the sequence). DNA42 = TTT GCG CCG AAA G*T*AAAA GTCCCC GTAAAA GTCCGC GGC TTT CCC GTAAAA. bThis measurement is for the second requirement for 2’-dCTP polymerization in DNAd2or RNA42 (underlined in the sequence).

nucleotide. Control experiments demonstrated that the absence of [y-3ZP]araATPincorporation into primers was not due to a lack of primase activity. To ascertain whether araAMP was located at the primer 3’-terminus, primers were again synthesized from [32P]ATP and [)H]araATP under conditions where pol a was inhibited with aphidicolin and DN&. After purification, primers were treated with RNase T2 and phosphodiesterase. Control experiments showed that these conditions were sufficient to completely degrade primers and that 3’-NMPs were not hydrolyzed to nucleosides. Cleavage of a phosphodiester bond by either enzyme generates a 3’-nucleotide and a compound with a free 5’-hydroxyl. Hence, if the araAMP is at the primer 3’-terminus, the hydrolytic product will be araA, while if it is at an internucleotide position, the product will be 3’araAMP. Prior to enzyme treatment, the primers remained near the origin on silica gel TLC (R, = 0-0.15). After treatment, 3000-fold, and with the identical all-DNA primer, pol a preferred 2’-dCTP by 4900-fold. Discrimination of pol a against ddATP was also measured on DNA4, and RN&,. In this case, however, a 2’-dCMP needed to be polymerized onto both primers for these measurements. Again, there was less discrimination against ddATP polymerization when the primer was RNA (plus one 2’-dCMP) rather than all-DNA (Table 111). However, whereas pol a discriminated 87-fold less effectively against ddCTP when the primer was all-RNA, pol a only discriminated 13-fold less effectively against the ddATP with the RNA (plus one 2’-dCMP) primer. Previous work has indicated that pol a polymerization of an araNMP results in termination of the growing DNA strand (Mikita & Beardsley, 1988; Reid et al., 1988). However, we observed that with an RNA primer, pol (Y can first polymerize multiple araATPs and then polymerize 2’-dATP (Kuchta & Willhelm, 1991). This suggested that with an RNA primer, incorporation of an araNMP might not normally result in chain termination. Surprisingly, we found significant amounts of further polymerization after incorporation of araCMP into either DNA42or RNA4*,while there was almost no polymerization after an araCMP into DN& (Figure 2). We therefore quantified the rate of 2’-dNTP polymerization onto primers containing either an araNMP or a 2’-dNMP at the primer 3’4erminus (Table IV).3 Rates were measured under a single set of steady-state conditions. In each case, the rate of polymerization onto an araNMP-terminated primer was much less than the corresponding 2’-dNMP-tenninated primer. There was a large variation in the rate of polymerization onto araCMP-terminated primers, indicating that the nature of the DNA and perhaps also the 2’-dNTP being polymerized affect the polymerization rate. In both cases examined, polymerization onto an araNMP-terminated RNA primer was significantly faster than polymerization onto the corresponding DNA primer (compare RNA42and DNA42). How Does the Primer 3’- Terminal Nucleotide Affect Interactions with Pol a? Incorporation of a 2’-dNTP analogue Polymerization rates onto araAMP/dAMP-terminatedprimers were only measured with the araAMP/dAMP at the first possible site where it can be incorporated.

Inhibition of DNA Primase and Pol a! by Arabinofuranosylnucleoside

Biochemistry, Vol. 31, No. 19, 1992 4125

Table V: KD for Nucleotide Binding to Various Pol anDNA Binary ComDlexes K D (PM) for nucleotide at primer 3’-terminus“ dTTP ddTTP 250 dCMP araCMP 90 1500 ddCMP 35 500 “The DNA used was DN& elongated by one nucleotide (DN&*). AraCMP and ddCMP were incorporated using KF (exo-), and 2’-dCMP was incorporated using pol a. DN&+c = TCC ATA AGG TAT TCA CAT C T* AGT GTA GAT cTT ~~

Scheme I EmDNA

XTP

11 1lNTP

E*DNA*XTP (i)

NTP

XTP

E*DNA*NTP e E*DNA*NTP*XTP (ii)

E*DNA*NTP*NTP

I

!

E*DNA*pppNpN

XTP

-

E*DNA*pppNpN*XTP

II

IlNTP

abcd

ef gh

i j kl

FIGURE2: Elongation of DNA,, DNA42,and RNA42 containing araCMP at the 3’-terminus. Primer strands were 5’-32P-labeled. Lane a is R N b 2 , and lane b is R N b 2after incubation with pol a and four dNTPs. Lane c shows RNA42after polymerization of an araCMP onto the 3’-terminus, and lane d shows the products after incubation of pol a and dATP, dGTP, and dTTP with the araCMP-elongated RNA42. Lanes e-h and i-1 correspond to lanes a-d, except the substrate was D N b 2 in lanes e-h and DNA, for lanes i-1. The major stop sites observed in lanes d and h correspond to the second site in each sequence where 2’-dCTP (araCTP) was required.

Table IV: Polymerization onto Template-Primers Containing either a 3’-Terminal AraNMP or 2’-dNMP rate with NMP at 2’-dNMP/ 3‘- termirate with template-primef nus rate (pmol m i d ) araNMP dAMP 5.3 41 DNA, araAMP 0.13 dCMP 1.5 >250 K0.006 araCMP DNA42 dAMP 2.6 90 araAMP 0.029 dCMP 2.4 114 araCMP 0.021 RNA42 dAMP 1.2 17 0.069 araAMP dCMP 0.74 8.4 araCMP 0.088 “Primers containing the appropriate 2’-dNMP or araNMP at the 3’-terminus were synthesized by incubating pol a, D N k , DNA42,or RNAs2and the appropriate dNTPs and araNTPs. ~

~~

into the growing DNA strand could alter the binding properties of the DNA. We measured the KD for the pol WDNA species generated after polymerization of 2’-dCTP, araCTP, or ddCTP onto DNAG. The KD for the DNA, with a 3’-terminal 2’-

E*DNA*pppNpN*NTP

J

E

I

E*DNA*pppNpNpX

+

DNA

+

pppNpN (iii)

-

+

XTP E

+

DNA

+

PPPNPNPX (iv)

further polymerization cycles

dCMP = 0.6 pM, similar to the KD for DNA, prior to addition of a 2’-dCTP [0.6pM (Sheaff et al., 1991)l. After polymerization of araCTP, KD = 1.5 pM, and after polymerization of ddCTP, KD = 1.3 pM. An additional mechanism by which incorporation of an araNMP or ddNMP could inhibit pol a! activity is via formation of a pol a-DNAdNTP ternary complex. Thus, binding of the next correct 2’-dNTP to pol cueDNA complexes was measured (Table V), where the DNA contained either a 5‘terminal araCMP or ddCMP. It is not possible to directly measure the KD for 2’-dTTP binding to the pol amDNA complex where the primer contains a 3’-terminal 2’-dCMP due to the rapid polymerization reaction. Thus, to estimate the effect of changing the 3’-terminal nucleotide from a 2’-dCMP to either araCMP or ddCMP, we determined the KD for ddTTP binding to each pol wDNA complex. The ddTTP bound only 2-fold weaker when the 3’-terminus was ddCMP rather than 2’-dCMP, suggesting that the 3’-hydroxyl on the primer is relatively unimportant for binding of the next correct 2’-dNTP. Additionally, as the primer terminus was changed from araCMP to ddCMP, the relative affinity for 2’-dTTP or ddTTP did not change.

D1scussloN Primase Inhibition. DNA primase is an essential enzyme for DNA replication and, as far as is known, is only involved in DNA replication. As such, it would be an ideal target for chemotherapeutics aimed at growing cells. Pol a! can only elongate primers at least seven nucleotides long; hence, primase inhibitors only need prevent primers from reaching this length.

4126 Biochemistry, Vol. 31, No. 19, 1992 As shown in Scheme I, NTP analogues (Le., XTP) could inhibit primase by several mechanisms, including the following: (i) the inhibitor competes with the initial NTP that will form the primer 5’-terminus; (ii) the inhibitor competes with the second, or latter, NTPs that are added to the growing primer, thereby decreasing the rate of primer synthesis by direct competition; (iii) the inhibitor binds to the primase-DNA. primer ternary complex, and allows dissociation of the DNA template and/or the primer; (iv) the inhibitor is incorporated into the growing primer strand, and upon incorporation terminates further polymerization. Within this framework, we examined several NTP analogues. For each compound tested, inhibition was reflected in a decreased total number of primers synthesized as well as an increased fraction of short primers. For those inhibitors not incorporated into primers, an increased fraction of short primers indicates that inhibition proceeds via mechanism iii while a decreased number of initiations could be due to either mechanism i or mechanism ii-we cannot discriminate between these two possibilities. Incorporation into primers with consequent termination (mechanism iv) is operative with araATP, 2‘-dATP, and probably araCTP. The 2’- and 3’-hydroxyls of the incoming NTP appear relatively unimportant for binding to primase. Inhibition due to just nucleotide binding to primase (mechanisms i-iii in Scheme I) only decreases 2-fold upon removal of the 2’hydroxyl (compare 3’-dATP and ddATP, and CTP with 2’dCTP; Table I), while removal of the 3’-hydroxyl had almost no effect (compare CTP with 3’-dCTP, and 2’-dCTP with ddCTP, Table I). Base complementarity also plays a relatively small role in inhibition of primase by NTPs. For those cases where the inhibitor was not incorporated into the primer (3’-dNTPs and ddNTPs), the noncomplementary cytosine analogue inhibited primase activity on poly(dT) only 7-fold less than the complementary adenosine analogue. The 3’-hydroxyl of the incoming nucleotide is critical for polymerization, since neither 3‘-dATP nor ddATP was polymerized. The 2’-hydroxyl is not essential since primase polymerized 2‘-dATP. A most surprising feature of primase, however, is its preference for nucleotides containing a 2’hydroxyl in the ara configuration, both for polymerization and for binding. With araATP, primase preferentially polymerized araATP by at least a factor of 4. Additionally, whereas primase did not incorporate CTP into primers on a poly(dT) template, small amounts of araCTP were incorporated. In addition to preferentially polymerizing araNTPs, primase likely also prefers binding nucleotides with a 2’-hydroxyl in the ara configuration. In primase assays measuring araCTP incorporation, less than 10% of primers contained araCMP under conditions that gave 52% inhibition, indicating that primer termination via araCTP polymerization can only account for 20% of the inhibition. The majority of primase inhibition by araCTP must have been due to simple binding of araCTP at one of the NTP binding sites. However, since araCTP is a 5-fold better inhibitor than CTP, araCTP must bind tighter than CTP. This also suggests that araATP binds better than ATP. Primase must contain two NTP binding sites in order to catalyze the initial polymerization event (2ATP pppApA), and araNTPs must bind to the site for incoming NTPs to account for their polymerization. However, it remains unknown if araNTPs can likewise bind to the site where the 5’-terminal NTP binds. While araATP is polymerized by primase, primase will not use araATP as the 5’-terminal nucleotide of the primer. This is consistent with the inability of primase to elongate primers

-

Kuchta et al. Scheme I1 XTP

E-DNA,

E-DNAaTP (i)

E*DNA*dNTP

E*DNA,,x

I/“

1

E*DNA,+

7

1 1“

E*DNA,+x*dNTP

E

+

DNA,+x

(ii)

slow ciii,E.DNAn+x+,j~

containing a 3’-terminal araAMP, since for the initial polymerization event, the NTP that will become the primer 5’-terminus is simultaneously the 3’-terminus of the “growing primer”. Similarly, 2’-dATP was not utilized as the 5’-terminal nucleotide. Unlike araAMP, however, 2‘-dAMP was found both at the primer 3’-terminus and at internucleotide positions. After incorporation of 2’-dAMP into primers, primase appeared to polymerize further ATPs onto the 2’-dAMP residue. An alternative possibility is that after primase incorporated 2’-dAMP into primers, pol a then polymerized further 2’dATPs, even though assays contained 200 pM aphidicolin to inhibit pol a. Using purified yeast primase, Brooks and Dumas (1989) also concluded that primase can incorporate 2’-dAMP into primers and then polymerize further ATPs. Pol a Inhibition. Potential mechanisms by which a nucleotide analogue (e.g., XTP) could inhibit pol a are summarized in Scheme I1 and include the following: (i) direct competition with the 2’-dNTP for binding to the pol a-DNA complex; (ii) incorporation of the inhibitor into the DNA and resultant chain termination; (iii) incorporation of the inhibitor into the DNA followed by binding of the next correct 2’-dNTP to generate a stable pol a.DNAIdNTP complex. Initially, we will only discuss inhibition where the primer is DNA. Pol a discriminated minimally between polymerization of araCTP or araATP versus the corresponding 2’-dNTP [(kc,,/ K~>d~~p/(k~t/Khl),raNTPl. Similarly, Parker et al. (1991) showed that pol a does not discriminate against 2-fluoroaraATP polymerization. This rapid polymerization of araNTPs indicates that simple competitive binding is relatively unimportant (mechanism i in Scheme 11), since the pol aDNA-araNTP complex will rapidly break down via either araNTP dissociation or polymerization. Incorporation of araATP or araCTP into DNA resulted in a DNA that was only slowly elongated by pol a (mechanisms ii and iii in Scheme 11), consistent with previous work (Mikita & Beardsley, 1988). The actual rate varied considerably as the DNA sequence was varied. That DNA sequence should cause such large changes in polymerization rates is consistent with studies on misincorporation of dNTPs by other DNA polymerases, where varying the DNA sequence can alter the misincorporation rate up to 1000-fold (Fersht et al., 1982; Kuchta et al., 1988; Lai & Beattie, 1988). The slow rate of elongation was not due to the araNMPcontaining DNA not binding to pol a. The KD for DNA containing a 3’4erminal araCMP (1.5 pM) was similar to the identical DNA containing a 3’-terminal 2’-dCMP (0.6 pM), indicating that the 3’-hydroxyl at the primer terminus is not required for binding. Similarly, a 3’-terminal ddCMP did not greatly alter binding to pol a ( K D = 1.3 pM). The presence of a ddNMP or araNMP at the primer terminus appears to have only small effects on the binding of the next correct 2’-dNTP to the pol asDNA complex. The KDs for 2’-dTTP or ddTTP binding to DNA, containing araCMP at the 3’-terminus (DN&+=,CMp) increased only 3-fold compared to DNAG containing ddCMP at the 3’4erminus

Inhibition of DNA Primase and Pol a by Arabinofuranosylnucleoside (DNhMdCMP). While we cannot directly measure the KD for 2’-dTTP binding to DNAG containing 2’-dCMP at the 3’terminus (DNAG+dCMP), we can estimate it to be ca. 17 pM. This estimate assumes that the ratio of the KDs for binding 2’-dTTP to DNAG+dCMP,DNAG+araCm and DN&+ddcMp will be the same as for ddTTP binding to these DNAs. The ratio for ddTTP binding to DNAG+araCMpor DNAG+,jdCMp was similar to the ratio of 2’-dTTP binding to DNAG+araCMpor DNAG+d,jCMp, suggesting it will also be true for ddTTP and 2’-dTTP binding to DNAG+dCMp and DNAG+ddCMpThus, removal of the 3’-hydroxyl from the primer terminus only modestly reduces binding of the next correct 2’-dNTP. RNA Primers. The pol aeprimase complex likely synthesizes at least the initial portion of Okazaki fragments in vivo; hence, pol a must frequently polymerize dNTPs onto RNA primers (Linn, 1991). We previously observed that on a poly(dT) template, pol a will polymerize multiple araAMPs onto a primase-synthesized oligo(A) primer (Kuchta & Willhelm, 1991), suggesting that pol a might interact differently with RNA primers than with DNA primers. Comparing identical RNA and DNA primer-templates (RNA42and DNA4&,we found that with the RNA primer, pol a discriminated 4-fold better against araCTP during the polymerization reaction. Conversely, we observed the opposite effect with ddCTP-pol a discriminated 85-fold less well against ddCTP polymerization when the primer was RNA rather than DNA. As dNMPs were polymerized onto the RNA primer, the decreased discrimination against ddNTP polymerization and enhanced discrimination against araCTP polymerization disappeared. Polymerization of just one 2’-dNMP onto RNA4, mitigated the effect of the RNA primer, suggesting that the effects of the RNA primer decrease rapidly as the primer 3’-terminus moves away from the last NMP. However, since the nucleotide polymerized directly onto the RNA primer is 2’-dCTP and the next nucleotide is 2’-dATP, we cannot exclude the possibility that the reduced effects also reflect the different nucleotides. For primers containing an araNMP at the 3’terminus, pol a used RNA primers more effectively than the identical DNA primers. Together, these data indicate that changing the primer from DNA to RNA results in large and unpredictable effects. Most data on purified pol a have been obtained on DNA primers, even though RNA primers are very important in vivo. For understanding how inhibitors function in vivo, particularly compounds such as ddATP or ddCTP that result in chain termination, the data with DNA primers potentially understate their effects on pol a. General Comments. The araNTPs have been extensively studied, yet the actual mechanism by which they inhibit DNA replication in vivo remains unclear. Kufe and =workers found that when cells are treated with arabinofuranosylnucleosides, araNMPs are incorporated into the DNA (Major et al., 1981, 1982), although only a small fraction of the araNMPs are found at the 3’-terminii of DNA. In vitro, however, incorporation of araNMPs into the DNA results in a species that pol a and pol 6 can only poorly elongate (Lee et al., 1980; Mikita & Beardsley, 1988; Reid et al., 1988). Pol a elongates RNA primers containing a 3’-terminal araNMP much more readily than the corresponding DNA primer; hence, the large amounts of araNMPs found at internucleotide linkages in vivo may reflect their incorporation onto RNA primers and subsequent elongation by 2’-dNTP polymerization. Additionally, in vivo there are likely a large number of other proteins associated with the replicative DNA polymerases which may further modulate their ability to elongate primers containing 3’4erminal araNMPs.

Biochemistry, Vol. 31, No. 19, 1992 4727

AraATP is a potent inhibitor of both pol a and primase, raising the question of which activity is the target in vivo. Inhibition of either activity could result in reduced rates of DNA synthesis. Recently, Catapano et al. (1991) concluded that inhibition of DNA replication by araCTP is due to DNA polymerase inhibition, while 2-fluoro-araATP inhibits primase. It is curious that two compounds so closely related inhibit DNA synthesis via different pathways in vivo. What makes RNA primers different? It is not solely due to a 2‘-hydroxyl at the 3’-terminus since effects of the RNA primer are detectable after polymerization of a 2’-dNMP or araNMP onto the primer. The different structures of a RNAmDNA duplex (A-helix) and DNA-DNA duplex (B-helix) could be important (Chou et al., 1989; Hall & McLaughlin, 1991; Reid et al., 1983). Alternatively, pol cy might specifically interact with 2’-hydroxyls of primer residues distant from the primer 3’-terminus. We are presently examining what makes RNA primers “different”. Polymerization of dNTPs onto RNA primers likely only occurs when cells are replicating. Since compounds can differentially inhibit pol a with RNA or DNA primers, polymerization onto RNA primers could be an attractive target for specifically inhibiting DNA replication with lesser effects on DNA repair. ACKNOWLEDGMENTS We thank Drs. Carlos Catalan0 and Donna Shewach for stimulating discussions during these studies and helpful comments on the manuscript. REFERENCES Brooks, M., & Dumas, L. B. (1989) J. Biol. Chem. 264, 3602-36 10. Catapano, C. V., Chandler, K. B., & Femandes, D. J. (1991) Cancer Res. 51, 1829-1835. Chou, S.-H., Flynn, P., & Reid, B. (1989) Biochemistry 28, 2435-2443. Dicioccio, R. A., & Srivastava, B. I. S . (1977) Eur. J . Biochem. 79, 41 1-418. Fersht, A. (1985) Enzyme Structure and Mechanism, 2nd ed., W. H. Freeman, New York. Fersht, A. R., Knill-Jones, J. W., & Tsui, W.-C. (1982) J . Mol. Biol. 156, 37-5 1. Hall, K. B., & McLaughlin, J. L. (1991) Biochemistry 30, 10606-1061 3. Hay, R. T., Hendrickson, E. A., & DePamphillis, M. L. (1984) J . Mol. Biol. 175, 131-157. Johnson, R. A,, & Walseth, T. F. (1979) Adv. Cyclic Nucleotide Res. 10, 135-167. Kaguni, L. S.,Rossignol, J.-M., Conaway, R. C., Banks, G. R., & Lehman, I. R. (1983) J. Biol. Chem. 258,9037-9039. Keeney, R. E., & Buchanan, R. A. (1975) Ann. N.Y. Acad. Sci. 255, 185-189. Kitani, T., Yoda, K.-Y., & Okazaki, T. (1984) Mol. Cell. Biol. 4, 1591-1596. Kuchta, R. D., & Willhelm, L. (1991) Biochemistry 30, 797-803. Kuchta, R. D., Mizrahi, V., Benkovic, P. A,, Johnson, K. A., & Benkovic, S . J. (1987) Biochemistry 26, 8410-8417. Kuchta, R. D., Benkovic, P., & Benkovic, S . J. (1988) Biochemistry 27, 67 16-6725. Kuchta, R. D., Reid, B., & Chang, L. M. S . (1990) J . Biol. Chem. 265, 16158-16165. Lai, M.-D., & Beattie, K. L. (1988) Biochemistry 27, 1722-1728. Larder, B. A., Darby, G., & Richman, D. D. (1989) Science 243, 1731-1734.

4128 Biochemistry, Vola31, No. 19, 1992 Lee, M. Y. W. T., Byrnes, J. J., Downey, K.M., & So, A. G.(1980)Biochemistry 19, 215-219. Linn, S.(1991) Cell 66, 185-187. Major, P. P., Egan, E. M., Beardsley, B. P., Minden, M. D., & Kufe, D. W. (1981)Proc. Natl. Acad. Sci. U.S.A. 78, 3235-3239. Major, P. P., Egan, E. M., Herick, D. J., & Kufe, D. W. (1982)Biochem. Pharmacol. 31, 2931-2944. Maniatis, T., Fritsch, E. F., & Sambrook, J. (1982)Molecular Cloning A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Mikita, T., & Beardsley, G. P. (1988) Biochemistry 27, 4698-4105.

Kuchta et al. Parker, W. B., Shaddix, S. C., Chang, C.-H., White, E. L., Rose, L. M., Brockman, R. W., Shortnacy, J. A., Secrist, J. A,, & Bennet, L. L. (1991)Cancer Res. 51,2386-2394. Reid, D. G.,Salisbury, S. A., Brown, T., Williams, D., Vasseur, J.-T., Raper, R. B., & Imbach, J. L. (1983)Eur. J. Biochem. 135, 307-314. Reid, R., Mar, E L , Huang, E&, & Topal, M. D. (1988) J. Biol. Chem. 263, 3898-3904. Sheaff, R., Ilsley, D., & Kuchta, R. D. (1991)Biochemistry 30, 8590-8597. Weil, M., Jacquillat, C., Auclerc, M. F., Achaison, G., Chartung, C., & Bernard, J. (1980)Recent Results Cancer Res. 80, 36-41.